The division of labor between and within countries is driven by two fundamental forces, comparative advantage and increasing returns. We set up a simple Ricardian model with a Marshallian input sharing mechanism to study their interplay. The key insight that emerges is that the interaction between agglomeration economies and comparative advantage involves a fundamental tension which is intricately affected by trade costs. A reduction of trade costs fosters the dispersive impact of comparative advantage in sectors governed by this force whilst the impact of agglomeration economies is enhanced by trade cost reductions in the increasing returns sector. The key implication for international trade is that the wage ratio between large and small economies is not only shaped by the primitives that determine agglomeration economies and comparative advantage but also, and differentially, by the sectoral levels of trade costs. The fundamental implication for an economic geography context where labor is mobile across locations is that partial agglomeration emerges when agglomeration economies are strong relative to comparative advantage, and this is more likely the lower are trade costs in increasing returns sectors and the higher are trade costs in sectors governed by comparative advantage. The model may serve as a foundation for an urban system where the endogenously emerging larger city exhibits more diversity in production.